Digitimes reports that Hynix developes VGA, 2MP, 3MP and 5MP sensors to utilize its 8" fabs. This should increase the market share of Korean CIS companies, including Samsung and SETi. Samsung has orders from first-tier clients Nokia, Dell and HP, while SETi has captured a nice slice of market in China. Hynix is offering more attractive prices to win orders over its competitors, according to Digitimes sources.
Image sensors are really on bargain sale. All gonna be losers... sorry
ReplyDeleteWe, image sensor designers, need to be more creative to generate the value. Otherwise, you are right, unfortunately.
ReplyDeletetoday I heard on the financial news that Kodak is on the watch list to default on bond payments over the next two quarters.
ReplyDeletethere doesn't seem to be anyplace to hide: the high volume products are priced so low that no profit is left, the high end products are too low volume to be worth introducing new products and the guys that make them are teetering on the brink of financial ruin.
In many ways the CMOS 'barbiecam' market looks like DRAM: lots of suppliers that aren't making a dime, but are running lots of wafers anyway.
It's not all that bad, like in DRAMs. On the low end there is wafer level camera technology which is supposed to make VGA camera even smaller and cheaper to manufacture than they are now. This should give an edge to the companies that are able to adapt it faster. BTW, there are no such a package competition in DRAMs.
ReplyDeleteOn the high megapixel handset end, there is a huge difference in image quality between the current crop of 1.4um pixels - nothing like this in DRAMs. Those companies that are able to provide the acceptable quality and not spent too much in development costs, might enjoy nice profits here.
So, not everything is lost on CIS market.
Aptina is really the question mark. Mircon can't sell it and runs it just for cash. Now it will lose money. Will Micron shut it down or keep it?
ReplyDeleteas soon as there are multiple suppliers of WLC slugging it out for the same sockets, why would there still be any remaining margin for the vendor? Usually the lowest cost supplier is the only one that makes a profit if anyone does. That's certainly true in DRAMs. Look at Samsung.
ReplyDeleteIn DRAMs and there is little value placed on exceeding the specifications needed to do the job: usually that just means the product cost was higher because the die was larger than necessary....
Unfortunately for some of the cellphone uses, what I hear is the pixel resolution is more of a checkbox item than anything else. This may be controversial: but my sales guys tell me that price and "megapixels" are more important than is image quality at the accounts they have.
> as soon as there are multiple suppliers of WLC slugging
ReplyDelete> it out for the same sockets, why would there still be
> any remaining margin for the vendor?
WLC can be improved even more: smaller pixels bring smaller camera size, integrated EDoF can solve process inaccuracies or, alternatively, process tolerances can be tightened, resolution can go up - that is, there is some room for differentiation to protect prices in each niche. At least this is my hope.
> Unfortunately for some of the cellphone uses, what I
> hear is the pixel resolution is more of a checkbox
> item than anything else.
Still, all the large cellphone manufacturers check image sensors quality and want the best. Most put quite high quality threshold which is not easy to pass.
This is a direct result of companies like Nokia trying to "commodotize" image sensors. In DRAM the specs are all preset and the only thing to compete on is price. Image sensors so far have not succumbed to full standardization pressure from large volume buyers but the trend is definitely there.
ReplyDeleteWhat is really needed is further consolidation in the market. Too many players right now, and thankfully China hasn't really started playing yet in the mid range market place.
-EF
Unfortunately the DRAM and FLASH markets need consolidation and it hasn't happened. I agree that the big camera phone guys want to see a commodity part so they can play one vendor against the other.
ReplyDeleteI expect that strategy to win and the end result to be a commoditization of the low end to mid range modules. The longer the world wide economy is in the tank the more likely I think this is to happen: too many suppliers chasing too little business.
The talks about commodotization and likening sensors to DRAMs started well before Nokia produced its first camera-phone. Since then the industry switched over to 4T pixels with few players lost, few others came, then the pixel shrunk to 1.4um, again not everybody succeeded, but few others joined the club and now there is a transition to even smaller pixel, BSI and WLC.
ReplyDeleteI'd expect the same pattern continue in the future with few more newcomers around the corner.
Sorry Image sensor, but I disagree. Before camera phones selling a million sensors was a big deal. CCDs or CMOS. Because of vertical integration, the Japanese image sensor companies did not have much commodization pressure and they were basically the only high volume image sensor manufacturers. With CMOS, the first big application was webcams and Logitech and Intel were the main players and perhaps Logitech tried this a bit, but it was the cell phone manufacturers, Nokia in particular, that really tried to drive towards standardization. This continues to today.
ReplyDelete-EF
I wonder how often EF disagrees with himself.
ReplyDeleteWeb is a great thing, it helps to refresh memories.
ReplyDeleteHere is the first Nokia camera-phone, available since Q2 2002:
http://en.wikipedia.org/wiki/Nokia_7650
Here is the talk about CMOS sensors commoditization, from January 2001:
http://www.siliconimaging.com/ARTICLES/CMOS%20Market.htm
http://www.ovt.com/pdfs/op_electronicBusiness_01_00_2001.pdf
"Long term, however, as CMOS sensors increasingly become a commodity, many current players are unlikely to survive."
The push for standardization and commodization started well before any camera phones made it to the public.
ReplyDeleteOnce the market opportunities became clear in the mid 1990's, I don't think it took great imagination to predict commoditization of a high volume semiconductor device or the other market force consequences like supplier consolidation, lower prices, etc.
The driver of this for the CMOS image sensor market was the camera phone and the most active of the manufacturers was and continues to be Nokia.
You might also be interested in this oldy but goody article from Sept 2000:
http://www.ericfossum.com/Papers/EE%20Times%20Sept%202000%20article.pdf
-EF
Note that your 11/3/09 17:45 comment is technically correct. But I thought you were trying to imply it was not related to the camera phone industry. This is what I disagree with.
ReplyDeleteIf you were just talking about timing, then you are probably right.
-EF